|Home | About | Journals | Submit | Contact Us | Français|
Activator protein 2α (AP-2α) is a putative tumor suppressor, and various reports have described the loss or reduction of AP-2α expression in cutaneous malignant melanomas, as well as in cancers of the prostate, breast and colon. Previously, AP-2α was shown to attenuate β-catenin/T-cell factor-4 (TCF-4) nuclear interactions and β-catenin/TCF-4-dependent transcriptional activity in human colorectal cancer cells [Q. Li, R.H. Dashwood, Activator protein 2alpha associates with adenomatous polyposis coli/beta-catenin and Inhibits beta-catenin/T-cell factor transcriptional activity in colorectal cancer cells, J. Biol. Chem. 279 (2004) 45669–45675]. Here, we show that in vivo gene delivery of AP-2α suppressed intestinal polyp formation in the Apcmin mouse, and protected against the development of anemia and splenomegaly. Immunoblot analyses and immunohistochemistry following gene delivery revealed an increase in AP-2α expression in the mouse intestinal mucosa and liver. Co-immunoprecipitation experiments provided evidence for interactions between AP-2α, β-catenin, and adenomatous polyposis coli (APC) proteins in mouse intestinal mucosa, as well as in a primary human colorectal cancer. Collectively, these studies support a tumor suppressor role for AP-2α in the gastrointestinal tract, and suggest that AP-2α represents a novel target for therapeutic intervention in human cancers characterized by dysregulated Wnt signaling.
AP-2α belongs to the AP-2 family of transcription factors, which are critical regulators of gene expression during vertebrate development, embryogenesis, and transformation [1–3]. AP-2α−/− mice die perinatally with cranio-abdominoschisis and severe dismorphogenesis of the face, skull, sensory organs, and cranial ganglia . In various human cancer cell lines, AP-2α over-expression is associated with inhibition of cell growth and induction of cell cycle arrest/apoptosis . Reduction or loss of AP-2α expression has been observed in cutaneous malignant melanoma and cancers of the breast, colon, and prostate [6–9], and AP-2α expression predicted survival outcomes in ovarian cancer patients .
We previously reported the novel finding that AP-2α forms a complex with adenomatous polyposis coli (APC)/β-catenin, attenuates β-catenin/T-cell factor-4 (TCF-4) nuclear interactions, and inhibits β-catenin-dependent reporter activity in human colorectal cancer cells . This suggested that AP-2α has the potential to interfere with β-catenin/TCF transcriptional activity, and that AP-2α might serve as a novel therapeutic target in cancers with dysregulated Wnt signaling . To test this hypothesis in vivo, we used a gene delivery system for AP-2α in the Apcmin mouse. This animal model is genetically predisposed to tumor formation in the gastrointestinal tract due to impaired function of the APC protein, resulting from a mutation that introduces a premature translational stop codon at amino acid 850 in one APC allele . The results of the present investigation showed that over-expression of AP-2α caused a highly significant suppression of spontaneous intestinal polyps, with evidence for AP-2α/APC/β-catenin interactions in vivo.
pcDNA3.1-AP-2α was generated by subcloning mouse cDNA-encoding full-length AP-2α into pcDNA3.1(+) (Invitrogen, Carlsbad, CA) between HindIII and EcoRV. APC fragments were generated from pCMV/APC using standard PCR-based methods, and PCR products were cloned into pcDNA3.1(+) . Plasmids were isolated using QIAGEN QIA filter Plasmid Maxi Kit (QIAGEN, Valencia, CA), with endotoxin removal using the MiraCLEAN® Endotoxin Removal Kit (Mirus, Madison, WI).
GST pull-down was performed as described previously . GST-β-catenin fusion protein was expressed in the BL21 strain of Escherichia coli and purified on glutathione–Sepharose 4B beads (Amersham Biosciences, Piscataway, NJ). [35S]-labeled AP-2α and fragments of APC were prepared using the TNT® quick coupled transcription/translation system (Promega, Madison, WI).
Male C57BL/6 J-Apcmin/+ (Apcmin) mice and C7BL/6 J+/+ (wild type) mice were obtained from the Jackson laboratory (Bar Harbor, Maine). pcDNA3.1-AP-2α was delivered by tail vein injection using the TransIT® In Vivo Gene Delivery System (Mirus Bio Corporation, Madison, WI). Immediately before injection, TransIT In Vivo Polymer Solution and endotoxin-free sterile water were mixed with pcDNA3.1-AP-2α, or with empty pcDNA3.1(+) as vector control, according to the manufacturer’s protocol. In a pilot immunohistochemistry study, mice were injected with a construct expressing myc-tag-His-AP-2α. The first injection (10 μg plasmid) was performed on mice at 7 weeks of age. In a follow-up study with intestinal polyps as the end-point, 3–4 mice per treatment were injected with vector or AP-2α expression construct, and the injections were repeated every other week for 2, 3, 4, or 5 five injections (i.e., 20, 30, 40, or 50 μg plasmid total). Twenty-four hours after the last injection, mice were sacrificed using an overdose of CO2, blood was drawn by cardiac puncture, and the hematocrit was recorded. The intestinal tract was divided into five sections (duodenum, jejunum, upper ileum, lower ileum, and colon), opened longitudinally, and examined by individuals blinded to the treatment status of the animal. A record was kept of the number and location of all polyps. Other tissues were weighed and snap-frozen in liquid nitrogen. All procedures were approved by the Institutional Animal Care and Use Committee.
Immunoblotting and co-immunoprecipitation experiments were performed as before . Whole cell lysates were prepared from human and mouse tissues by homogenizing each sample in ice-cold lysis buffer (20 mM Tris– HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerol phosphate, 1 mM Na3VO4, 1 μg/ml leupeptin) and centrifuging at 4 °C, 15,000 rpm, for 15 min. Proteins were separated on 4–12% bis–tris gels (Novex, San Diego, CA) or on 3–8% Tris–acetate gels (for APC) and transferred to nitrocellulose membranes, followed by immunoblotting with primary antibody and horseradish peroxidase-conjugated secondary antibody. Primary antibodies were rabbit polyclonal anti-AP-2α (Santa Cruz Biotechnology, Santa Cruz, CA), mouse monoclonal anti-β-catenin (Transduction Labs, Lexington, KY), rabbit polyclonal anti-Cyclin D1 (Neomarker, Fremont, CA), and mouse monoclonal anti-β-actin (Sigma–Aldrich, St. Louis, MO). Image analysis and quantication was performed on an AlphaImager™ 2200 system (AlphaInnotech Corporation, San Leandro, CA).
Formalin-fixed, paraffin-embedded specimens of liver and small intestine were sectioned at 4–5 μm and placed on Microprobe slides (Fisher Scientific, Pittsburgh, PA). Following rehydration, slides underwent high temperature antigen retrieval using citrate buffer (pH 6.0; DakoCytomation, Carpentaria, CA) for 10 min in a microwave pressure cooker (Tendercooker) and placed at room temperature for 20 min. Slides were washed in Automation buffer (Biomeda, Foster City, CA) followed by blocking in 3% H2O2 in methanol for 10 min. Immunohistochemical staining was performed as reported before . The primary antibody was rabbit polyclonal anti-AP-2α (Santa Cruz Biotechnology), used at a dilution of 1:1000 and applied for 30 min at room temperature. Universal Negative Control Rabbit (DakoCytomation) was used a negative control.
Formalin-fixed, paraffin-embedded specimens of liver, small intestine, and spleen were sectioned at 4–5 μm and stained with Hematoxylin and Eosin. Cytology (paint-brush) preparations of bone marrow were stained with Wrights-Giemsa. All slides were examined by a board certified anatomic pathologist, blinded to the treatment groups.
Results were expressed as mean ± SD. Data for test versus control was compared using Student’s t-test. As shown in the figures, results were considered as statistically significant at the *P < 0.05 or **P < 0.01 level of significance, respectively.
We demonstrated previously  that in human colorectal cancer cells the Basic region of AP-2α binds to the heptad and armadillo repeats of APC (Fig. 1A). To confirm that APC is an intermediate between AP-2α and β-catenin, we first attempted to pull-down 35S-labeled AP-2α using GST-tagged β-catenin, in the presence and absence of full-length APC or APC fragments. AP-2α was pulled-down by GST-β-catenin in the presence of full-length APC (Fig. 1B), but not by AP-2α alone, indicating that APC is essential for AP-2α/APC/β-catenin complex formation. We generated APC fragments containing the N-terminal, middle, and C-terminal portions of the APC protein, and designated these as nAPC, mAPC and cAPC, respectively (Fig. 1A). nAPC represents the AP-2α binding region containing the Heptad and Armadillo (Arm) repeats, mAPC has the 15-amino acid (15-aa) and 20-aa repeats associated with β-catenin binding, and cAPC contains the Basic region that binds neither AP-2α nor β-catenin. β-Catenin failed to pull-down AP-2α in the presence of any of these APC fragments (Fig. 1B). Thus, the ability of AP-2α to sequester β-catenin away from TCF is APC-dependent, and requires full-length APC for a functional AP-2α/APC/β-catenin complex.
To test the tumor suppressor function of AP-2α in vivo, we used a gene delivery system in a mouse model of dysregulated Wnt signaling . Thus, AP-2α was introduced into Apcmin and wild type mice via tail vein injection, using a nonviral approach for gene delivery. The basic methodology was reported before [15–17], including as means of expressing human APC in Apcmin mice . The commercial in vivo gene delivery system used here was designed primarily for distribution to the liver, but we sought to test whether it might target other organs, such as the small intestine.
We first introduced myc-His-AP-2α into Apcmin mice to demonstrate that exogenous AP-2α can be over-expressed in vivo. By design, this short duration pilot study did not include polyps as an end-point. Immunoblotting with anti-myc-tag antibody revealed strong expression levels of myc-His-AP-2α in the livers of mice injected with pcDNA3.1-myc-His-AP-2α (Fig. 2A). As expected, no exogenous AP-2α was observed in the vector controls (Fig. 2A). Immunohistochemical staining revealed that, following gene delivery, AP-2α was over-expressed in the cytoplasm and/or nuclei of hepatocytes (Fig. 2B). Rather than uniform immunostaining, AP-2α was strongly expressed in individual cells throughout the liver. No immunostaining was detected in the liver of vector controls (Fig. 2C). Similar observations were made in the small intestine; rather than uniform immunostaining, individual epithelial cells along the villus expressed high levels of AP-2α after gene delivery (Fig. 2D). This pattern was only obvious at high magnification in the mice injected with AP-2α expressing construct, and immunostaining was weak to undetectable in the villi of vector controls (Fig. 2E).
In vivo gene delivery had no obvious deleterious effect on animal behavior, or on food and water consumption. Histopathological examination of spleens, livers, and small intestines, and cytological examination of bone marrows, revealed no differences between the treatment groups. Collectively, these findings provided initial confirmation that the methodology was a quick, simple approach for in vivo delivery of AP-2α.
In subsequent experiments, Apcmin or wild type mice were injected with pcDNA3.1-AP-2α expression construct, or with pcDNA3.1(+) as the vector control, and intestinal polyps were assessed as in our prior studies [19–21]. For both wild type and Apcmin mice, there were no deleterious effects of the treatment regimen on organ weights or final body weights, and no macroscopically visible tumors were observed in any of the wild type mice (data not presented).
In Apcmin mice, however, there was an inverse association between final tumor multiplicity and AP-2α dose administered (Fig. 3A). Maximal tumor suppression was attained after three injections (30 μg plasmid), and one or two additional injections had no further inhibitory effect. Given the similar outcomes for mice injected with 30, 40, or 50 μg plasmid (Fig. 3A), the data were combined and compared with controls given vector alone. As shown in Fig. 3B, AP-2α treatment resulted in a highly significant suppression of tumor multiplicity; vector controls had an average of 50.2 ± 11.2 polyps, whereas animals injected with the AP-2α expressing construct had 23.0 ± 2.0 polyps (P < 0.01). The distribution of polyps along the gastrointestinal tract was as reported before [19–22], with the majority of tumors occurring in the ileum (Fig. 3C). AP-2α treatment suppressed polyps most effectively in the upper and lower ileum (P = 0.005 and 0.046, respectively). Polyps also appeared to be slightly smaller in size for mice injected with the AP-2α expressing construct compared with mice given vector alone (not shown).
In addition to spontaneous polyps throughout the gastrointestinal tract, the Apcmin mouse has a phenotype that includes development of anemia and splenomegaly. Previously, we observed an inverse association between the hematocrit (packed cell volume) and the multiplicity of polyps in Apcmin mice . In the present study, hematocrits were 22.7 ± 3.07 and 53.6 ± 2.56 in vector-treated Apcmin and vector-treated wild type mice, respectively (open bars, Fig. 3D). AP-2α treatment had no effect on hematocrit in wild type mice, but there was a significant improvement in Apcmin mice (37.8 ± 2.05, P = 0.001 compared with Apcmin mice treated with vector). Spleens in Apcmin mice were grossly enlarged, with an average organ weight of 0.25 ± 0.045 g versus 0.063 ± 0.006 g for vector controls (open bars, Fig. 3E). AP-2α treatment decreased the average spleen size in Apcmin mice to 0.13 ± 0.017 g (P = 0.02), and this reduction was evident by gross observation of spleens (Fig. 3F). Thus, AP-2α protected against splenomegaly and anemia in the Apcmin mouse.
Western blotting revealed low levels of AP-2α in the ileum of vector-treated Apcmin mice, but high levels of AP-2α were detected following in vivo gene delivery (Fig. 4A). Immunohistochemistry studies of polyps revealed generalized, low-level expression of AP-2α throughout the tumor tissue (data not presented). By immunoblotting, tumors typically had lower expression of AP-2α than the adjacent normal-looking tissue, but this was not always the case (Fig. 4B). As reported before in Apcmin mice , β-catenin and the β-catenin/Tcf downstream target Cyclin D1 were highly over-expressed in tumors compared with adjacent normal-looking tissue (Fig. 4B).
Co-immunoprecipitation experiments with tissue lysates from the intestines of Apcmin mice established that full-length Apc was pulled-down by anti-AP-2α and anti-β-catenin antibodies, whereas anti-APC and anti-β-catenin antibodies immunoprecipitated AP-2α (Fig. 4C, left). We also provide initial evidence for AP-2α/APC/β-catenin interactions in a primary human colorectal cancer; β-catenin antibody immunoprecipitated AP-2α, and AP-2α antibody pulled down full-length APC (Fig. 4C, right).
The main goal of this investigation was to extend the prior work which implicated AP-2α as a possible tumor suppressor. Inasmuch as in vivo gene delivery of AP-2α inhibited intestinal polyps in the Apcmin mouse, and protected against the development of anemia and splenomegaly, we can conclude that our findings are consistent with a tumor suppressor role for this transcription factor. With appropriate modification of the in vivo gene delivery protocol, such as shortening the duration between injections, it might be feasible to inhibit polyp formation to a greater extent than observed here, although this possibility awaits experimental verification.
Consistent with its proposed tumor suppressor function, AP-2αexpression tended to be lower in polyps than adjacent normal-looking tissue, although this was not universally true. By the time polyps arise in Apcmin mice the adjacent normal-looking tissue is scarcely “normal”, and microadenomas or other precursor lesions might conceivably affect AP-2α levels. Moreover, immunoblotting and immunohistochemical studies might not provide the complete picture. Inhumancolorectal cancer cells transfected with exogenous AP-2α, immunoblotting of nuclear extracts revealed high levels of AP-2α without changes in β-catenin or TCF-4; however, co-immunoprecipitation experiments revealed that β-catenin/TCF-4 nuclear interactions were attenuated, whereas AP-2αβ-catenin/APC interactions had increased . We extended these observations and showed, for the first time in vivo, that AP-2α/APC/β-catenin interactions were detected in the intestinal mucosa of Apcmin mice and in a primary human colorectal cancer.
We postulate that early stages of colorectal cancer might benefit most from therapy targeted at AP-2α, because the AP-2α/APC/β-catenin complex requires at least one full-length copy of APC (Fig. 1B), and in late-stage colon cancers loss of heterozygosity can delete both wild type APC alleles . APC is known to shuttle in and out of the nucleus ; thus, it is possible that nuclear APC, along with AP-2α, sequesters β-catenin away from TCF transcription factors as a means of controlling β-catenin/TCF signaling. Consistent with this paradigm, polyps with low levels of AP-2α expression had high levels of Cyclin D1, a down-stream β-catenin/TCF target. However, it is premature to conclude from the present investigation that AP-2α/APC/β-catenin complex formation was the only (or even primary) mechanism for tumor suppression. For example, Cyclin D1 and other downstream β-catenin/TCF targets might be influenced by AP-2α as a transcriptional regulator. Additional studies are warranted to clarify the precise mechanisms involved. The current preclinical model and in vivo gene delivery approach has advantages in terms of simplicity and speed, but it is not ideal in that AP-2α was targeted primarily to liver. A conditional model targeting AP-2α to the intestinal and/or colonic mucosa might provide more definitive insights into the role of AP-2α as a tumor suppressor in the gastrointestinal tract.
In summary, we have shown for the first time that AP-2α suppresses the development of intestinal tumors arising spontaneously in a mouse model of dysregulated Wnt signaling. The findings are consistent with studies in primary human colon cancer and cultured colorectal cancer cells indicating that AP-2α associates with APC/β-catenin, with the potential to disrupt β-catenin/TCF interactions. Further work is needed to corroborate whether this mechanism might provide a therapeutic avenue in clinical cases of colorectal cancer.
We thank B. Vogelstein (Johns Hopkins University, Baltimore, MD) for pCMV-APC, H. Clevers (Hubrecht Institute, The Netherlands) for pcDNAI-β-catenin, and Steven F. Moss (Rhode Island Hospital, Providence, RI) for supplying specimens of primary human colon cancer. G.A. Orner, W.M. Dashwood, V. Elias, M. Myzak, D. Yu, and R. Wang kindly assisted in various aspects of the animal studies, and K.A. Fischer provided technical support for the immunohistochemistry and histology work. This work was supported by NIH Grants CA090890, CA65525, and CA122959, and by the Cell Imaging and Analysis Core of the Environmental Health Sciences Center, funded by NIEHS center Grant P30 ES00210.
Conflict of interest statement
The authors declare that no financial or personal relationships with other people or organizations inappropriately influenced the present work.